unofficial mirror of guile-user@gnu.org 
 help / color / mirror / Atom feed
From: Taylan Kammer <taylan.kammer@gmail.com>
To: "Dr. Arne Babenhauserheide" <arne_bab@web.de>,
	Guile User Mailing List <guile-user@gnu.org>
Subject: Re: Syntax-Case macro that selects the N-th element from a list
Date: Mon, 5 Apr 2021 17:08:29 +0200	[thread overview]
Message-ID: <7a5af132-13a2-65eb-5a36-f1e4e6c840ba@gmail.com> (raw)
In-Reply-To: <87im5155pu.fsf@web.de>

On 05.04.2021 13:30, Dr. Arne Babenhauserheide wrote:
> Hi,
> 
> In dryads-wake I need selection of the element in a list in a macro from
> user-input. Currently I have multiple macros, and the correct one (which
> strips the non-selected choices) is selected in a simple cond:
> 
> (define-syntax-rule (Choose resp . choices)
>    "Ask questions, apply consequences"
>    (cond
>     ((equal? resp 1) ;; resp is user-input. It is a natural number.
>      (Respond1 choices))
>     ((equal? resp 2)
>      (Respond2 choices))
>     ((equal? resp 3)
>      (Respond3 choices))
>     (else
>      #f)))
> 
> For this however I have three syntax-case macros:
> 
> (define-syntax Respond1
>   (lambda (x)
>     (syntax-case x ()
>       ((_ ((question consequences ...) choices ...))
>         #`(begin
>            (respond consequences ...)))
>       ((_ (choices ...))
>         #`(begin #f)))))
> 
> (define-syntax Respond2
>   (lambda (x)
>     (syntax-case x ()
>       ((_ (choice choices ...))
>         #`(begin
>            (Respond1 (choices ...))))
>       ((_ (choices ...))
>         #`(begin #f)))))
> 
> (define-syntax Respond3
>   (lambda (x)
>     (syntax-case x ()
>       ((_ (a b choices ...))
>         #`(Respond1 (choices ...)))
>       ((_ (choices ...))
>         #`(begin #f)))))
> 
> 
> I would like to get rid of those three definitions and replace them by
> at most two (one that strips N initial list entries, and Respond1).
> 
> I cannot move to procedures, because I have code that must be executed
> only during final processing, and when I evaluate any of the
> consequences (as it happens with procedure-arguments), then the timing
> of the code execution does not match anymore. So I must absolutely do
> this in macros.
> 
> 
> I’ve tried to get that working, but all my tries failed. Is there a way
> and can you show it to me?
> 
> This is a minimal working example. The output should stay the same,
> except for part 4, which needs this change to work (see at the bottom),
> but I would like to:
> 
> - replace Respond2 and Respond3 by something recursive, so resp can have
>   arbitrary high values (not infinite: max the length of the options) and
> - replace the cond-clause by a call to the recursive macro.
> 
> (define-syntax-rule (respond consequence consequence2 ...)
>   (begin
>     (write consequence)
>     (when (not (null? '(consequence2 ...)))
>       (write (car (cdr (car `(consequence2 ...))))))))
> 
> (define-syntax Respond1
>   (lambda (x)
>     (syntax-case x ()
>       ((_ ((question consequences ...) choices ...))
>         #`(begin
>            (respond consequences ...)))
>       ((_ (choices ...))
>         #`(begin #f)))))
> 
> (define-syntax Respond2
>   (lambda (x)
>     (syntax-case x ()
>       ((_ (choice choices ...))
>         #`(begin
>            (Respond1 (choices ...))))
>       ((_ (choices ...))
>         #`(begin #f)))))
> 
> (define-syntax Respond3
>   (lambda (x)
>     (syntax-case x ()
>       ((_ (a b choices ...))
>         #`(Respond1 (choices ...)))
>       ((_ (choices ...))
>         #`(begin #f)))))
> 
> 
> (define-syntax-rule (Choose resp . choices)
>    "Ask questions, apply consequences"
>    (cond
>     ((equal? resp 1)
>      (Respond1 choices))
>     ((equal? resp 2)
>      (Respond2 choices))
>     ((equal? resp 3)
>      (Respond3 choices))
>     (else
>      #f)))
> 
> 
> (display "Choose 1: should be bar:")
> (Choose 1 (foo 'bar) (foo 'war 'har) (foo 'mar) (foo 'tar))
> (newline)
> (display "Choose 2: should be warhar:")
> (Choose 2 (foo 'bar) (foo 'war 'har) (foo 'mar) (foo 'tar))
> (newline)
> (display "Choose 3: should be mar:")
> (Choose 3 (foo 'bar) (foo 'war 'har) (foo 'mar) (foo 'tar))
> (newline)
> (display "Choose 4: should be tar:")
> (Choose 4 (foo 'bar) (foo 'war 'har) (foo 'mar) (foo 'tar))
> (newline)
> (display "Choose 5: should be #f:")
> (Choose 5 (foo 'bar) (foo 'war 'har) (foo 'mar) (foo 'tar))
> (newline)
> 
> 
> Best wishes,
> Arne
> 

Is there a reason you want to separate 'choose' from the various
'respondN' macros?  That seems superfluous to me.

And if the number of choices is limited, you don't even need to use
procedural macros.  The following works:

;; Note: I like using the notation <foo> for template variables.
;; Feels highly intuitive to me.

(define-syntax choose
  (syntax-rules ()
    ((_ 1 <c1> <rest> ...)
     (respond <c1>))
    ((_ 2 <c1> <c2> <r> ...)
     (respond <c2>))
    ((_ 3 <c1> <c2> <c3> <r> ...)
     (respond <c3>))
    ((_ 4 <c1> <c2> <c3> <c4> <r> ...)
     (respond <c4>))
    ((_ 5 <c1> <c2> <c3> <c4> <c5> <r> ...)
     (respond <c5>))
    ((_ <r> ...) #f)))

You could also leave out the final clause and thus let it be an implicit
syntax error, or explicitly invoke 'syntax-error' with an explanation.

But if you want to support an arbitrary numbers of choices, the
following procedural macro will do it:

(define-syntax choose
  (lambda (stx)
    (syntax-case stx ()
      ((_ <n> . <choices>)
       (let* ((n (syntax->datum #'<n>))
              (choice-list (syntax->datum #'<choices>))
              (choice-datum (list-ref choice-list (- n 1)))
              (choice-stx (datum->syntax #'<choices> choice-datum)))
         #`(respond #,choice-stx))))))

I've written a very trivial explanation of syntax-case usage here, in
case anyone struggles as I did at first:

http://taylanub.github.io/doc/syntax-case-example.scm.txt


Hope that helps!

- Taylan



      parent reply	other threads:[~2021-04-05 15:08 UTC|newest]

Thread overview: 6+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-04-05 11:30 Syntax-Case macro that selects the N-th element from a list Dr. Arne Babenhauserheide
2021-04-05 12:21 ` Linus Björnstam
2021-04-05 13:40   ` Linus Björnstam
2021-04-05 16:24     ` Dr. Arne Babenhauserheide
2021-04-05 13:51 ` Tim Van den Langenbergh
2021-04-05 15:08 ` Taylan Kammer [this message]

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

  List information: https://www.gnu.org/software/guile/

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=7a5af132-13a2-65eb-5a36-f1e4e6c840ba@gmail.com \
    --to=taylan.kammer@gmail.com \
    --cc=arne_bab@web.de \
    --cc=guile-user@gnu.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).